DETAILED SURFACE ANALYSIS OF INCREMENTAL CENTRIFUGALBARREL POLISHING (CBP) OF SINGLE-CRYSTAL NIOBIUM

SAMPLES∗

A. D. Palczewski† , H. Tian, O. Trofimova, and C. E. ReeceSRF Institute, Jefferson Lab, 12000 Jefferson Ave. Newport News, VA 23606, USA

Abstract

We performed Centrifugal Barrel Polishing (CBP) onsingle crystal niobium samples/coupons housed in a stain-less steel sample holder following the polishing recipe de-veloped at Fermi Lab (FNAL) in 2011 [1]. Post CBP, thesample coupons were analyzed for surface roughness, crys-tal composition and structure, and particle contamination.Following the initial analysis each coupon was high pres-sure rinsed (HRP) and analyzed for the effectiveness ofcontamination removal. We were able to obtain the mirrorlike surface finish after the final stage of tumbling, althoughsome defects and embedded particles remained. In addi-tion, standard HPR appears to have little effect on remov-ing embedded particles which remain after each tumblingstep, although final polishing media removal was partiallyaffected by standard/extended HPR.

INTRODUCTION

Superconducting radio-frequency (SRF) cavity technol-ogy is one of the key technology driving current particleaccelerators. Modern SRF cavities are made out of high pu-rity niobium sheets which are then formed into cavities. Toachieve state–of–the art performance, these cavities mustgo through a multi-step process which condition the in-ner surface to support a high Q and maximum accelerationgradient (EAcc). The usual method to treat the inner sur-face included initial bulk buffered chemical polish (BCP)and/or electropolishing (EP), followed by a high tempera-ture bake to remove hydrogen, followed by another lightBCP/EP, high pressure rinse (HPR), and low bake. Thistechnique can achieve Qo > 2 × 1011 and EAcc > 40MV

mfor a ILC 1.3 GHz cavity at 2K [4].One of the alternate techniques to reduce bulk and light

chemistry is mechanical polishing, pioneered at KEK inJapan [2]. In addition to reducing chemistry, recent workat FNAL by Cooper et. al [1] has shown that centrifugalbarrel polishing (CBP) can reduce the surface roughnessby a order of magnitude lower than chemistry alone [3];which in turn possibly raises the Q thereby improving the

∗Authored by Jefferson Science Associates, LLC under U.S. DOEContract No. DE-AC05-06OR23177. The U.S. Government retains anon-exclusive, paid-up, irrevocable, world-wide license to publish or re-produce this manuscript for U.S. Government purposes.

† ari@jlab.org

Figure 1: Sample mounting of niobium coupon in CBPholder. Left, the stainless steel holder - diameter is 9inches; middle, coupon mounted on a 2 3

4 conflate flangebefore CBP, the hole in the flange contains a set screw toadjust the height of the rod; right, coupon mounted insideSS container - looking though the opposite coupon holderport.

performance.

Most CBP recipes rely on cavity testing with trial anderror to ascertain a recipe for optimal performance. Whilethis technique works, there should be a physical quantitywhich can help guide further recipe refinement. TheseCBP quality factor could include visual appearance, sur-face roughness, contamination size and shape, and/or crys-tal structure deformation. There are two readily availableways of ascertaining this information; one is to CBP a cav-ity until it reaches a “good” performance and then sacrificethe cavity by cutting out samples which can then be placedin a analysis equipment such as atomic force microscopy(AFM), scanning election microscopy (SEM), and electronback scatter diffraction (EBSD); the other would be to per-form the analysis on sample housed in a host sample holder.While the first may be ideal, the second is cheaper andthroughput can be much higher.

In this work, we perform CBP on single crystal nio-bium samples/coupon housed in a stainless steel couponholder following the mirror smooth CBP recipe developedat FNAL, and analyzed the coupons with multiple surfacecharacterizing techniques. We were able to reproduce themirror finish published yet our mirror finish also containedscratches and embedded media. Our results indicate thateach step in the process removes all signs (embedded me-dia) of the previous steps, yet residual embedded mediafrom each step always remain. The effects of high pres-sure rinsing were also evaluated with promising results forcomplete colloidal silica removal with multiple/extended

THPO071 Proceedings of SRF2011, Chicago, IL USA

908 04 Material studies

Table 1: FNAL recipe comparison of removal rates on a fine grain cavities and large grain coupon removal rates followingthe same recipe at JLab. The spread in our removal rates comes from the variation between coupons.

FNAL recipe processing time (hours) published removal rates (µmh ) coupon removal rates (µmh )

9mm x 9mm triangles (Course) 10 11 27-31RG-22 cones (Medium) 12 3 7-9400 mesh alumina (Polish 1) 15 >1 1-1.5800 mesh alumina (Polish 2) 20 >1 0.5-1colonial silica (Polish 3) 40 negligible negligible

Figure 2: Image of a representative coupon for each CBP process start from the PRE-CBP BCP sample on the left andmoving right, 9mm×9mm KM + 40:1 DI water:TS Compound, RG-22 cones + 40:1 DI water:TS Compound, 400 meshalumina + wood blocks + DI water, 800 mesh alumina + wood blocks + DI water, and colonial silica (40nm) + woodblocks.

rinsing cycles.

CENTRIFUGAL BARREL POLISHING(CBP)

All single crystal samples were CBP’ed inside a stain-less steel cavity holder designed by Peter Kneisel at JLab[5], andmounted into a CBP machinemanufactured specif-ically for SRF polishing by Mass Finishing [6]. Wemounted each 1 3

8 coupon on a niobium rod using high tem-perature mount wax (South Bay Technology Quick Stick135 mounting wax) fitted into a 2 3

4 conflate flange, see Fig-ure 1 middle. Because the holder has been used in the pastwith smaller coupons the edges of the SS holder have beenwarn down, leaving an exposed edge (Figure 1 right) whichdeviates from the curvature of the coupon. Each couponwas cut out of single crystal piece of niobium with a sur-face orientation of 110, verified by EBSD.

Following the FNAL recipe, each step was performedwith the sample holder filled 50% with media and thentopped off with liquid (DI water + TS compound or DI wa-ter only or colidal silica). The details of the recipe andremoval rates comparison are shown if Table 1, with theimage of the coupon after each step shown in Figure 2. Be-tween each CBP step the coupons were rinsed with acetoneto remove the mounting wax. This allowed the removalrates to be measured at the center of the sample using anOlympus Panametric NDT 25DL Plus ultras sonic thick-ness gage. Since the coupon holder contained 4 samples,all samples went through the first two step, and then foreach addition step a coupon was removed; leaving only onecoupon for the final step.

COUPON ANALYSIS

Scanning Electron Microscopy (SEM)

SEM was used to find the contamination level and em-bedded media for each step of CBP. For the first step usingKM triangles (Al2O3 and SiO2) both about 8-10 µm in di-ameter were left behind. Step two left behind SiO2 about1-3 µm in diameter, in addition Al2O3 was also found (notshown), suggesting it was left behind from the first step (orcontamination). Step three and four left behind Al2O3 inthe range of 3-5 µm. The last step was unable to remove allthe Al2O3 from the third or forth step. Example embeddedmedia indicative from each step is show in Figure 3. Col-loidal silica is only found on our sample in crack/fisherson the surface which contain Al2O3 particles, or where itlooked like a peace of Al2O3 used to be embedded (seesection on HPR).

Atomic Force Microscopy (AFM)

Each couponwas analyzed by AFM overmultiple 50 µm× 50 µm section to find the RMS height and Zmax (highestpoint above the RMS height). The compiled data is shownin Figure 4. One can see the first step creates a surfacerougher than the BCP surface, yet after the medium CBPstep the RMS surface roughness and Zmax are recovered.Two surprising feature of the data is polishing step 1 (4from Figure 4) may not be needed, and the manufacturedmedia would appear to have a more uniform finish than thefirst two polishing steps.

Proceedings of SRF2011, Chicago, IL USA THPO071

04 Material studies 909

Figure 3: SEM image of embedded media from each CBPstep moving - from top to bottom: course, medium, polish1, polish 2, polish 3. Particle size and description found intext.

Electron Back Scatter Diffraction (EBSD)

Each coupon was analyzed with EBSD to ascertain thesurface crystal structure for all CBP step. The initial BCPsurface showed a clear 110 crystal orientation (averageconfidence index C.I. = 0.87), but every step of CBPdestroyed the crystal orientation, at least within the 40nmprobing depth of EBSD (Figure 5). Only after the fullrecipe did any coupon show any sign of crystal structurewith EBSD, although still not nearly as periodic as the BCPsurface which showed a C.I. = 0.87 (CBP C.I. = 0.20).

High Pressure Rinsing (HPR)

After each polishing step, the coupon underwent a clean-ing process of a low pressure DI water rinse, ultrasonic

Figure 4: AFM data for each step in the CBP process. RMSheight taken over multiple a 50m X 50 m section of eachcoupon, the error bars represent the span of measurementsfrom multiple sampling areas. Zmax for each coupon, theerror bars represent the span of measurements from multi-ple sampling areas.

Figure 5: Electron back scatter diffraction (EBSD) beforeCBP on a fresh BCP surface (left) and after the FNAL CBPrecipe (middle). The gray background is the SEM image ofthe surface, the color insert (right) indicates the appearanceof a periodic crystal structure crystal orientation is presentat the surface. BCP C.I.=0.87 and CBP C.I.=0.20

cleaning in acetone for 10 minute, ultrasonic cleaning inDI water with Micro–90 detergent for 10 minute, and afinal ultrasonic cleaning in DI water 10 minute. The ini-tial cleaning was unable to remove any embedded media,see previous sections. After the initial surface analysis, allcoupon underwent a high pressure (HPR) rinse in DI wa-ter at 1250 PSI and at a distance of 3.5-3.75 inches for 1hour to evaluate contamination removal. The HPR systemsetup used the standard single cell rinsing cycle for Jeffer-son laboratory where the coupons/cavity rotates around thespay head in a 360 degree pattern at 2 rpm, and the risingwand moves up at a stepping rate of 0.5 inches per minutefor a total of 80 passes, about twice as long as a standardsingle cell. The HPR was unable to remove the embeddedmedia from the first four steps, but it did seem to remove atleast some portion of the colloidal silica (Figure 6 left). Af-ter an additional 6 hours HPR, in the same way as before,more colloidal silica was removed (Figure 6 right). Wherethe HPR removed the colloidal silica small residue circlesremained, presumably (but not verified) from the glycerol

THPO071 Proceedings of SRF2011, Chicago, IL USA

910 04 Material studies

suspension, but this was also reduced by extended HPR.

Figure 6: SEM image surface CBP coupon after a 1 hourHPR cycle (left) and an additional 6 hour HPR (right), seetext for HPR information.

CONCLUSION

We have presented a systematic study of the CBP surfaceof single crystal niobium coupon using the 2011 FNALrecipe. We found the removal rates were 2-3 time higherthan what was published for standard fine grain niobiumcavities. This is possibly due to the samples being largegrain rather than the published fine grain, or a property ofthe stainless steel sample holder. We found all tumblingsteps left behind media embed which were not removedby HPR, but in the final polishing step the residual mediacould at least be reduced. In addition, it appears the firstpolishing step could be removed from the recipe, withouteffecting the final surface roughness.

REFERENCES

[1] C. A. Cooper and L. D. Cooley,http://lss.fnal.gov/archive/2011/pub/fermilab-pub-11-032-td.pdf, (May 31, 2011).

[2] T. Hiuchi, K. Saito, S. Noguch, M. Ono, E. Kako, T. Shishido,Y. Funahashi, H. Inoue and T. Suzuki, Proceedings of theWorkshop on RF Superconductivity 95, SRF95F34 734, Gif-sur-Yvette, France, (1995).

[3] H. Tian and C. Reece, Phys Rev Special Topics Accel.Beams, 13 083502, (2009).

[4] The International Linear Collider: A Technical ProgressReport, http://ilcdoc.linearcollider.org/record/32863, (July2011).

[5] P. Kenisal, TESLA Technology Collaboration (TTC), April19-22, Fermi national Lab, Chicago, IL, USA (2010).

[6] C. A. Cooper et al, Proc. of SRF2009, THPPO076, 806,Berline, Germany 806, (2009).

Proceedings of SRF2011, Chicago, IL USA THPO071

04 Material studies 911